... now with 35% more arrogance!

Tuesday, February 11, 2014

Random Dungeon Musings

Random Dungeon Generation is the new emerging topic this month, it seems.

There’s a post at the Tao of D&D about dissatisfying aspects of the random dungeon generator tables in the back of the DMG. There’s also a brief mention of random dungeon generation and stocking at Dreams in the Lich House, more of a musing on the necessity of random maps and tables to keep a mega-dungeon project from requiring enormous amounts of time to prepare.

I’ve thought a lot about it myself. And I’ll say first that half of what Alexis at Tao of D&D is not really a big deal. He thinks a major problem with the DMG generator, and other generators he’s seen, is that they produce too many empty rooms. No need to delve into the philosophy of empty rooms and how best to use them; Alexis is simply wrong about this being a major problem because it’s a trivial fix. Swap “Empty” with one of the other results, like “Monster”; problem solved. Although it would be better to change it to “Resource Room with chance of hidden vermin”. You want to leave breathing room for players to rest, or decision points for the expedition (search this room to see if there’s something useful, or not?)

Now, the other point Alexis made is more reasonable: random dungeon generation tends to produce less structure than you may like. It would make more sense for the dungeon to be separated into zones, divided by obstacles or secret doors or other restrictions to movement, Each zone would have its own list of monsters; only vermin or wildlife would be likely to spill over across zone boundaries, while major predators and intelligent denizens would be restricted to specific areas. You’d feel a natural flow, that way: as soon as you made your way across the Underworld Rapids, you’d notice fewer Devil Goblins and more Spiked Frogmen.

Now, I’ve had many ideas about random dungeons and have many others I’m mulling over, but I’d like to point out that the megaduungeon modules I posted a couple days ago are sort of built on that concept, and point the way to a solution: build each zone around a centerpiece room with special features and several generic room types for surrounding areas. If you are using random generators for solitaire play, roll the zone type first as you enter an area, then assign your vermin monster(s) and build a basic corridor structure. Exploring rooms is likely to reveal more monster types, more resource types, and eventually, you will find the “central” room that defines the theme of the area.

Your first table is going to look something like:
  1. Caverns+Lair
  2. Warrens
  3. Ruins+Lair
  4. Ruins+Curse
  5. Tombs
  6. Tombs+Hideout
  7. Caverns+Hideout
  8. Ruins+Hideout
  9. Caverns+Tribe
  10. Ruins+Tribe  
  11. Fortification
Etc. The idea being that each entry is a pair of “general structure” + “type of major threat”. There would be separate tables for each structure type: caverns, warrens (tightly-packed and dug by a beast,) ruins (crumbling, repurposed constructed area,) tombs (sort of warrens for the dead,) and forts (newer structures in good repair.) Each table would define the major active feature + minor room types, and the minor rooms would cluster around the major room. The zone table would also direct you to separate monster tables to roll up vermin, wild beasts, monstrosities, and inhabitants, with “Lairs” favoring beasts and other monstrosities as the dominant inhabitant, “Cursed” pretty much guaranteeing a monstrosity with an unusual ability and possible semi-intelligence, “Hideout” being a bandit camp, “Tribe” generating humanoid living areas, and “Fortification” having a more organized intelligent monster dominating lesser minions.

But I suppose I’ll have to develop this in more detail later.
Written with StackEdit.


  1. Wasn't that generator originally meant for solo play? The problems aren't as surprising when one bears that in mind. Levels need to be planned; designed. Randomness is great for stocking all the little things to make management easier (and what ref doesn't like a few surprises from something other than their players? "What is behind that door?"), but the input of a referee is absolutely necessary to produce a truly great dungeon. Even if one's fine with a crazy, structureless dungeon, someone needs to design those traps!

    Unrelated, but I had a thought earlier today regarding variable damage vs. the use of modifiers according to weapon and armor types, and just had to share it: the latter's too combat focused. If we're going to embrace the idea of D&D as an exploration game with monsters being a mere (albeit common) obstacle, shouldn't we consider potential environmental interactions when designing weapons? Doors, locks, treasure chests and even hewn stone walls are all common features that players can interact with, possibly via their weapons or spells. Variable weapon damage - and possibly differing damage types - would be great for that

    1. It does cover solo play, but that is what Alexis @ Tao of D&D is focusing on. He was saying he wouldn't be able to bear exploring a dungeon created during play using the DMG or most other RDG methods because they create too many empty rooms and not enough of a uniform theme in each area.

    2. I thought he said he wanted to use it as a tool for the referee; not for playing alone

      It's vital that certain aspects of a dungeon, including theming, be planned. The empty rooms are a more legitimate issue(if you don't like them, that is), though (as you noted) trivial to fix

  2. I believe my statements were directed to an unwillingness on the part of dungeon generation systems I've seen to expunge empty space from their construct - or at least suggesting that it is an endemic characteristic, rather than suggesting it is something that is difficult to solve.

    1. I try to avoid concentrating too much on speculating about the motivations of others, because even though it's *fun*, it's not very productive. Better to focus on whether a situation can be changed.

      That being said, I think the main problem with RDGs used for solitaire play is that empty rooms are empty of *everything*, not just inhabitants, traps, and treasure. Basicaally, tables should start at the top with the most obvious characteristics -- room shape and exits -- and zero in on finer and finer details: obvious contents of room, not so obvious, and finally hidden features.

      So, you get:

      1. Rectangular room, 30x50 aligned E/W, doors in E and W walls.
      2. Storeroom full of stacked wooden crates.
      3. Rats emerge from crates if disturbed.
      4. Crates contain iron rations (3d6-7 are still good) and rat's nest with 3d6 x20 gold coins, 2d6-6 gems of base 50 gp value.
      5. Secret door hidden in N wall, must move crates to search.

      When generating during play, what the player does should determine how much of that list you generate. Thus, a room that appears empty except for crates could mean much more, if you risk a search.